Return-Path: Delivered-To: apmail-new-httpd-archive@apache.org Received: (qmail 60791 invoked by uid 500); 28 Mar 2000 09:19:25 -0000 Mailing-List: contact new-httpd-help@apache.org; run by ezmlm Precedence: bulk X-No-Archive: yes Reply-To: new-httpd@apache.org list-help: list-unsubscribe: list-post: Delivered-To: mailing list new-httpd@apache.org Received: (qmail 60647 invoked from network); 28 Mar 2000 09:19:23 -0000 Message-ID: <009b01bf9896$5a58f940$0a1aa8c0@jetnet.co.uk> From: "David Reid" To: , References: <200003280153.LAA29842@silk.apana.org.au> <200003280236.VAA01302@k5.localdomain> Subject: Re: cvs commit: apache-2.0/src/lib/apr/dso/os2 Makefile.in dso.c dso.h Date: Tue, 28 Mar 2000 10:15:39 +0100 MIME-Version: 1.0 Content-Type: text/plain; charset="Windows-1252" Content-Transfer-Encoding: 7bit X-Priority: 3 X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook Express 5.00.2919.6600 X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6600 X-Spam-Rating: locus.apache.org 1.6.2 0/1000/N I have to admit that when I was looking for a problem with opening a file having the error string would have been very helpful. Perhaps Manoj's idea of a get_error routine for every type is the best one? It should make finding the error easier and should be easy enough to implement. I know Ryan doesn't like it though! d. ----- Original Message ----- From: "Jeff Trawick" To: Sent: Tuesday, March 28, 2000 3:36 AM Subject: Re: cvs commit: apache-2.0/src/lib/apr/dso/os2 Makefile.in dso.c dso.h > > One other thing, the OS/2 API call DosLoadModule() returns a valuable > > piece of information on failure, the name of the module that failed to > > load. When loading a DLL, that DLL may depend on a whole chain of other > > DLLs, any one of which may fail. Knowing which one did can save a whole > > lotta time. > > > > In 1.3, this information is put in the static error string for later > > retrieval by ap_os_dso_error(). For APR I'm thinking that it could be > > stored in the dso_handle_t for later retrieval by a suitable accessor > > function. > > There are plenty of other situations where a particular os has special > information which, if available, greatly simplifies problem > determination. OS/390, for example, has a secondary error indicator > for every system call which provides more information about the > error. In many cases it is able to pinpoint the error. This is > particularly helpful in the case of configuration problems. > > We could invent ways particular to different data types for saving > this information; we could add code to our favorite APR app for > checking to see if any such information was available. > > An alternate idea (shot down very quickly a few weeks ago) would be > for an APR app to give APR a function to call to record error > information. This dispenses with some issues completely, such as how > to store the error info in the object for later retrieval and when the > APR app should query the object for such info (all this multiplied by > the number of objects that should support this type of information). > > One platform-independent aspect of the trace is the use of a text > string to record the diagnostic information. The text string could be > used on any platform. If get-last-error functions are added to > various objects, they could return a text string for passing to > ap_log_error(). > > -- > Jeff Trawick | trawick@ibm.net | PGP public key at web site: > http://www.geocities.com/SiliconValley/Park/9289/ > Born in Roswell... married an alien... >